Avikal Sagar Email & Phone Number

Chicago, Illinois, United States

1. Project: eRoadMAP (eroadmap.epri.com)• Cleaning, processing and analyzing vehicle telematics and/or transportation electrification datasets provided by OEMs and fleet operators• Generating load forecasts at granular levels on a nationwide basis due to upcoming EVs2. Project: Custom Utility Dashboards• Processing large sets of EV registration/sales and VIO data• Creating dashboards for member utilities to help them visualize these numbers and other metrics (such as YoY growth) on maps within their specific territories and states

New York, New York, United States

• Developed a Python-based time-cost optimization tool for EV charging infrastructure using SciPy’s MILP solver• Leveraged data from 12 medium-duty EVs, 5 charger types, 3700 utility rate structures and load profiles as optimization parameters• Rendered the tool as a web application using HTML/CSS & Flask, with AWS as the host, for users to make sustainable charging decisions

Minneapolis, Minnesota, United States

• Extracted over 1.2 million values of spatiotemporal agroclimatological & soil moisture data from Copernicus satellite's ERA5 database, weather stations and IoT devices, to create the training set for predicting soil moisture changes (in VWC %) on turfgrass• Processed the dataset by removing outliers, imputing values to fill in null spaces, aggregating each variable on an hourly basis, and applying feature engineering techniques such as standardization at required places• Built Random Forest models (fine-tuned using Grid search & Randomized search), and LSTMs with RMSE scores in the range of 0.6-0.85• Deployed best-fit models to provide 12 or 24-hour predictions of soil moisture as a function of the forecasted weather & agroclimatological conditions• Models and their predictions were integrated on to a software UI built on PowerBI using Python extensions and plug-ins, to make it interactive and easy-to-visualize for end users such as turf managers and superintendents• Interactive features of the UI included simulating changes in forecasted moisture values by adding pre-determined amounts of irrigation at certain hours of the day that helped understand (a) whether irrigation is actually required or not, and (b) if needed, approximate time and maximum amount of irrigation to be applied - thus helping them conserve water while also maintaining turf health & playability

New York, New York, United States

Worked for Prof. Brügger's research group, in collaboration with Carleton Laboratory (Columbia University) and Oak Ridge National Laboratory (Tennessee), in implementing Machine Learning and Genetic Algorithms for the quantification of internal bridge cable mechanics.

• Performed time series forecasting of temperature and turbidity due to global warming for the upcoming five years in river Ganga at and around Varanasi, India by utilizing Machine Learning and Remote Sensing techniques• Converted the geographical data acquired (in .kml format) from Google Earth Engine to shapefiles (in .shp format) on QGIS and stored the shape, location, and attributes of the geographic features in the vector data format• Retrieved the minimum, maximum, and mean temperatures of the air 2 meters above the river, for each of these shapefiles, from the 'ERA-5' satellite on GEE code editor, and analyzed the non-stationarity & seasonality of the dataset• Deployed models such as Seasonal Autoregressive Integrated Moving Average (SARIMA) and Prophet for predicting air temperature above the river for the next five years with an RMSPE of 3.2% and implemented Recurrent Neural Networks such as LSTM to analyze river temperature, and subsequently, turbidity and nighttime radiance values for years 2022 and 2025• Predicted average river temperature showed an increase of 2 C and 2.5 C for city and non-city river stretches, respectively, in 2025 as compared to 2022. Within the same period, river turbidity and nighttime radiance also depicted an upsurge

• Identified and analyzed four deep eutectic solvents (ionic liquids) for carbon capture in post-combustion industrial processes• Created a correlation matrix and the values of solubility as a function of 10+ features such as structural parameters, molecular complexity, number of rotatable bonds, molecular weight, density, temperature and pressure were obtained• Built an Artificial Neural Network model to predict the solubility of CO2 at various temperature and pressure with an RMSE of 0.0032 and R-squared value of 0.9528, and also derived the relation between solubility, alkyl chain lengths and thermodynamic parameters

Master Of Science - Ms, Engineering Mechanics (Track: Computational & Data-Driven Engineering Mechanics), 3.8/4.0

Courses: 1. Data Science for Mechanical Systems 2. Neural Networks & Deep Learning 3. Principles & Applications of Sensors in Structural.

Bachelor Of Technology - Btech, Civil Engineering (Major), Information Technology (Minor), Major: 8.66/10, Minor: 9.00/10

Activities and Societies: Indian Society for Technical Education NITK Surathkal, 180 Degrees Consulting NITK

Class Xii - Physics, Chemistry And Mathematics, 94.00%